INDEX

Explanations

words related to updates and news

oai_token-act-pair · gpt-3.5-turbo

references to updates or revisions in content

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.1.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Negative Logits

bors

-0.80

stood

-0.78

stem

-0.74

adra

-0.74

thing

-0.71

filled

-0.71

bour

-0.70

powers

-0.70

ften

-0.69

lo

-0.69

POSITIVE LOGITS

Updated

1.13

CLASSIFIED

0.91

Posted

0.88

UPDATE

0.84

 millenn

0.83

 undermin

0.81

 Updates

0.81

 Update

0.80

Update

0.78

 UPDATE

0.77

Activations Density 0.005%

words related to updates and news

references to updates or revisions in content

No Known Activations